Changeset 27063

Timestamp:

01/22/10 18:40:27 (3 years ago)

Author:

rande

Message:

[sfSolrPlugin] add query analyser/guesser

Files:

• plugins/sfSolrPlugin/branches/sf1.2/lib/util/sfLuceneCriteria.class.php (modified) (4 diffs)
• plugins/sfSolrPlugin/branches/sf1.2/test/unit/util/sfLuceneCriteriaTest.php (modified) (2 diffs)

plugins/sfSolrPlugin/branches/sf1.2/lib/util/sfLuceneCriteria.class.php

@@ -37 +37 @@
-    TYPE_NONE = '',
-    TYPE_AND  = 'AND',
-
+
+
-
-  
@@ -177 +178 @@
-  }
-  
+  public function guessParts($phrase)
+  {
+    // initialize variable
+     $phrase = trim($phrase);
+     $phrase = str_replace('""', '', $phrase);
+     $phrase = str_replace('\'\'', '', $phrase);
+
+     $default_separators = array(' ', ',');
+     $phrase_separators  = array('\'', '"');
+     $separators = array_merge($default_separators, $phrase_separators);
+     $separator = $default_separators;
+
+     $current_phrase = "";
+     $contains       = 'contains';
+
+     $parts = array(
+       'must_contains' => array(),
+       'must_not_contains' => array(),
+       'contains' => array(),
+     );
+
+     for($i = 0; $i < strlen($phrase); $i++)
+     {
+       $char = $phrase{$i};
+
+
+       if(strlen($current_phrase) == 0)
+       {
+
+         if($char == '-')
+         {
+           $contains = 'must_not_contains';
+           continue;
+         } 
+         elseif($char == '+')
+         {
+           $contains = 'must_contains';
+           continue;
+         }
+
+         if(in_array($char, $phrase_separators))
+         {
+           $separator = array($char);
+           continue;
+         } 
+         else if(in_array($char, $default_separators))
+         {
+           $separator = $default_separators;
+
+           continue;
+         }
+       }
+
+       // end of the phrase
+       if(strlen($current_phrase) > 0 && (in_array($char, $separator) || strlen($phrase) - 1 == $i))
+       {
+
+         if(strlen($phrase) - 1 == $i && !in_array($char, $separator))
+         {
+           $current_phrase .= $char;
+         }
+
+         $parts[$contains][] = $current_phrase;
+
+         // restore default values
+         $contains       = 'contains';
+         $current_phrase = '';
+         $separator      = $default_separators;
+
+         continue;
+       }
+
+       $current_phrase .= $char;
+     }
+     
+     return $parts;
+  }
+  /**
+   * This method try to parse the provided $phrase into a valid solr query
+   * The method can handle +/- and quote grouping
+   *
+   * @param string $phrase 
+   * @return sfLuceneCriteria 
+   */
+  public function addPhraseGuess($full_phrase)
+  {
+
+    foreach($this->guessParts($full_phrase) as $section => $phrases)
+    {
+      if(count($phrases) == 0)
+      {
+        
+        continue;
+      }
+
+      $inner_type = ($section == 'must_contains' || $section == 'must_not_contains') ? 'AND' : 'OR';
+      $sign       = ($section == 'must_contains' ? '+' : ($section == 'must_not_contains' ? '-' : ''));
+      
+      $c = new sfLuceneCriteria;
+      foreach($phrases as $phrase)
+      {
+        $c->addPhrase($phrase, $inner_type);
+      }
+      
+      $this->add($sign.'('.$c->getQuery().')', 'AND', true); 
+    }
+    
+    return $this;
+  }
+  
+  /**
+   * This method try to parse the provided $phrase into a valid solr query
+   * The method can handle +/- and quote grouping
+   *
+   * @param string $field 
+   * @param string $phrase 
+   * @return sfLuceneCriteria 
+   */
+  public function addPhraseFieldGuess($field, $full_phrase, $type = sfLuceneCriteria::TYPE_AND)
+  {
+
+    $main_criteria = new sfLuceneCriteria;
+    foreach($this->guessParts($full_phrase) as $section => $phrases)
+    {
+      if(count($phrases) == 0)
+      {
+        
+        continue;
+      }
+
+      $inner_type = ($section == 'must_contains' || $section == 'must_not_contains') ? 'AND' : 'OR';
+      $sign       = ($section == 'must_contains' ? '+' : ($section == 'must_not_contains' ? '-' : ''));
+      
+      $c = new sfLuceneCriteria;
+      foreach($phrases as $phrase)
+      {
+        $c->addPhrase($phrase, $inner_type);
+      }
+      
+      $main_criteria->add($sign.'('.$c->getQuery().')', 'AND', true); 
+    }
+    
+    $this->addField($field, $main_criteria, $type, true);
+    
+    return $this;
+  }
+  
-  public function addField($field, $query, $type = sfLuceneCriteria::TYPE_AND, $force = false)
-  {
@@ -273 +421 @@
-   * @return sfLuceneCriteria
-   */
-
+, $clever = false
-  {
-    
@@ -611 +759 @@
-  public static function sanitize($keyword)
-  {
-
+
-    return sfLuceneApacheSolrService::phrase($keyword);
-  }

plugins/sfSolrPlugin/branches/sf1.2/test/unit/util/sfLuceneCriteriaTest.php

@@ -87 +87 @@
-$s = $criteria->getQuery();
-
-\\"
+
-
-try {
@@ -169 +169 @@
-$t->cmp_ok($s, '===', $expected, '->add() with empty string');
-
+$t->diag('testing addPhraseGuess()');
+
+$s = inst()->addPhraseGuess('Thomas -"zend framework"')->getQuery();
+$expected = '-("zend framework") AND ("Thomas")';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+$s = inst()->addPhraseGuess('"Thomas"   -"zend framework"')->getQuery();
+$expected = '-("zend framework") AND ("Thomas")';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+$s = inst()->addPhraseGuess('"Thomas"   -.zend')->getQuery();
+$expected = '-(".zend") AND ("Thomas")';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+$s = inst()->addPhraseGuess('Thomas Rabaix +"symfony expert" -"zend framework" +javascript -.net')->getQuery();
+$expected = '+("symfony expert" AND "javascript") AND -("zend framework" AND ".net") AND ("Thomas" OR "Rabaix")';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+$s = inst()->addPhraseGuess('Thomas Rabaix +"sym"fony expert" -"zen-d framework" +javascript -.net')->getQuery();
+$expected = '+("sym" AND "javascript") AND -("zen-d framework" AND ".net") AND ("Thomas" OR "Rabaix" OR "fony" OR "expert")';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+$s = inst()->addPhraseFieldGuess('name', 'Thomas Rabaix +"sym"fony expert" -"zen-d framework" +javascript -.net')->getQuery();
+$expected = 'name:((+("sym" AND "javascript") AND -("zen-d framework" AND ".net") AND ("Thomas" OR "Rabaix" OR "fony" OR "expert")))';
+$t->cmp_ok($s, '===', $expected, '->addPhraseGuess()');
+
+
+